@Ramana 

Thanks for the thoughtful feedback and for sharing your experience with LASIO and custom handling of variable mnemonics. You’re absolutely right that with highly variable LAS structures, especially when dealing with hundreds or thousands of potential curves, locking into a fixed wide schema can be inefficient. In those cases, storing the data in an “unpivoted” format can help accommodate curve variability while maintaining query flexibility.

We haven’t yet completed formal production-scale performance testing with millions of LAS files, but the approach is designed to scale with cloud-based ingestion. For workloads where long-term operational stability is a priority, future options like the Key Information Extraction agent brick may offer more production-ready capabilities and built-in optimizations.

It’s great to hear that you’re considering testing this method in your environment. I’d be very interested in learning how it performs in your specific workloads once you've tried it.